Working Paper No. 587 Reprisals Remembered: German-Greek Conflict and Car Sales During the Euro Crisis Vasiliki Fouka | Hans-Joachim Voth August 2016 Reprisals Remembered: German–Greek Conflict and ∗ Car Sales During the Euro Crisis Vasiliki Foukay Hans-Joachim Vothz August 2016 Abstract Limited attention and selective memory are key behavioral factors identified in the literature on cognitive biases and economic outcomes. We investigate how events trigger selective recall and thus change economic behavior. Following public disagreement between German and Greek politicians, Greek consumers drastically reduced their purchases of German automobiles – especially in areas affected by German reprisals during World War II. In response to contemporary political disagreements, Greeks living in areas where German troops committed massacres curtailed their purchases of German cars to a greater extent than those elsewhere. Current events can reactivate past memories, having a first-order effect on purchasing behavior. JEL Codes: D12, D74, F14, N14, N44 Keywords: consumer boycott, memory, political conflict, car sales, Euro crisis, German– Greek relations ∗For helpful suggestions we thank Alexander Apostolides, Leo Bursztyn, Ray Fisman, Nicola Gen- naioli, Yannis Ioannides, Tim Leunig, Guy Michaels, Stelios Michalopoulos, Nathan Nunn, Sonal Pandya, Elias Papaioannou, Luigi Pascali, Giacomo Ponzetto, and Nico Voigtlaender. Seminar participants at CREI, UPF, Harvard, the EREH-London conference, and the 12th Conference on Research on Economic Theory and Econometrics (Naxos) provided useful advice. The Hellenic Statistical Authority kindly pro- vided car registration data. Voth received funding via ERC-AdG. [email protected] [email protected] 1 The past is never dead. It’s not even past. —William Faulkner 1 Introduction Theoretical research in behavioral economics posits that limited attention, fallible mem- ory, and salience all affect individual decision making.1 Models of limited attention typ- ically predict underreaction to news, and there is substantial empirical support for this prediction. Investors, for example, often react slowly to earnings news released on Fridays or on days when many other companies are reporting (DellaVigna and Pollet, 2009; Hirsh- leifer et al., 2009); similarly, consumers change their behavior only gradually in response to the rankings of hospitals and colleges (Pope, 2009). Attention effects might play a role in the delayed adoption of new production techniques in developing countries (Duflo et al., 2011; Hanna et al., 2014).2 Also, savings can be increased by reminding people about the importance of retirement savings (Karlan et al., 2016). Complementary re- search in macroeconomics has explored the implications of inattention for business cycles (Mankiw and Reis, 2002; Sims, 2003). A closely related theoretical literature on memory limitations leads to the opposite prediction – namely, that news can lead to sharp changes in behavior if it “brings back” memories of past events (Mullainathan, 2002).3 Such remembrance is more likely to occur if the current situation is similar to the past one, a scenario that encourages recall or (in the language of cognitive psychology) “associativeness”. Some results in experimental psychology bear out the predictions of limited-memory models (Serman and Kim, 2002), yet there is no compelling evidence for any appreciable changes in economic behavior that are due to a suddenly remembered past. There is also only limited evidence that either limited attention or limited memory can have a significant effect on aggregate economic outcomes.4 1Simon(1955); Kahneman and Tversky(1982); Akerlof(1991); Gabaix et al. (2006); Köszegi and Szeidl(2013); Schwartzstein(2014). 2Another prominent finding is that excise taxes, which are included in the final price, reduce demand more than sales taxes, which are normally charged only at checkout (Chetty et al., 2009). 3Salience-based models such as those offered by Gennaioli and Shleifer(2010) and Bordalo et al. (2015) combine limited attention with limited memory; these models can accommodate both under- and overreactions. 4One important exception are Malmendier and Nagel(2011) and Malmendier and Nagel(2016), who show that long-lingering memories of major economic upheavals such as the Great Depression or high inflation in the 1970s can influence expectations both at the individual and the aggregate level, and affect 2 In this paper, we examine the interaction between contemporary events and local memory in a recent setting with high stakes for the agents involved: the sales of German cars in Greece during the sovereign debt crisis of 2010–2014. We show how local memory affected reactions to news, leading to much larger changes in some areas than in others. Before Greece could receive bailout loans from the European Union, it was required to implement stringent austerity measures. The German government in particular pushed for these belt-tightening measures. German newspapers were quick to blame “lazy South- erners” in Greece and elsewhere for their travails. The proposed policies created public discord between the German and Greek governments, and memories of past violent con- flict quickly resurfaced. During the occupation of Greece in 1941–1944, the German armed forces committed numerous war crimes, including mass executions of civilians; Oc- cupation policies also triggered a major famine (Mazower, 1995). As public exchanges between German and Greek politicians grew increasingly acrimonious after 2010, Greek demonstrators waved placards of Angela Merkel in Nazi uniform while protesting against the rise of a “Fourth Reich” in the form of EU-imposed austerity. In addition, consumer groups called for a boycott of German products. We focus on car purchases because cars are an iconic German product.5 Such pur- chases also represent a major investment for the average consumer. To pin down the importance of associativeness of memory, we focus on within-country differences in reac- tions to a public event: the sudden eruption of German–Greek animosity. In their bid to suppress partisan activity, German occupying forces carried out numerous reprisals that often involved burning whole villages and killing the entire (male) civilian population in the vicinity of partisan attacks (Mazower, 1995). The number and intensity of German massacres varied within Greece. We collect detailed geographical data on the location of German massacres during the Occupation, 1941–1944. To measure the severity of these attacks at the local level, we use the 1940 share of a prefecture’s population living in towns affected by massacres. For that purpose we rely on lists drawn up by the Greek govern- ment designating localities as “martyred towns”; that designation is based on criteria that include the percentage of homes destroyed and the loss of human life. The locations of these martyred towns are then matched to prefecture-level car registration data. We also compile an index of German–Greek political clashes based on articles in the overall asset returns. Relatedly, Koudijs and Voth(2016, forthcoming) show that personal experience in lending to stock market speculators shifted the Amsterdam stock market’s aggregate risk-bearing capacity in the 1770s. 5Some car manufactures use idiomatic and hard-to-pronounce German slogans in their advertising abro ad so as to polish their German credentials (as when Audi proclaimed “Vorsprung durch Technik” [advancement through technology] in English-speaking countries). 3 leading Greek newspaper, Kathimerini. This index identifies an upsurge of conflict after 2010, with several periods of extremely high public animosity. Public clashes between the German and Greek governments sharply reduced overall sales of German automobiles. Yet this effect was not uniform across Greece: in some areas, there was hardly any change in sales; elsewhere, a large decline was observed in the German market share. We find that the greater the number and intensity of massacres in a Greek prefecture, the greater the decline in the German market share. Figure1 illustrates the timing and magnitude of effects. This figure plots the share of articles devoted to contemporary German–Greek conflict alongside the difference – between reprisal and non-reprisal prefectures – in the market share of German autos. Until the outbreak of EU-related conflict, the average difference in German market share between reprisal and non-reprisal towns fluctuated around 0. Thereafter, this gap became large and negative, growing at an especially rapid pace during the period of maximum animosity just after the Greek default in 2012. As the German Finance Minister and the Greek President publicly traded insults, the difference (between reprisal and non-reprisal prefectures) in the market share of German cars grew to more than 1 percentage point (approximately two standard deviations); it then took a year for that difference to vanish. The effect is both large and significant. Once we take correlates into account, the effect of a “conflict month” on the German car market share in areas with massacres becomes even larger – rising to 3.4 percentage points. Since the pre-crisis market share for German manufacturers was 23 percent, it follows that the decline represents a persuasion rate of 15 percent.6 This value represents the additional decline of German market share in massacre provinces, and the effect is large. Only five of the 24 studies surveyed by DellaVigna and Gentzkow(2009) have similar or higher persuasion